Crash Detection Research Articles

An Empirical Study on How Sapienz Achieves Coverage and Crash Detection

AbstractSeveral tools for automatically testing Android applications have been proposed. In particular, Sapienz is a search‐based tool that has been recently deployed in an industrial setting. Although it has been shown that Sapienz outperforms several state‐of‐the‐art tools, it is still to be seen what features of SAPIENZ impact the most on its effectiveness. We conducted an extensive empirical study where we compare the impact of the search algorithm and the usage of motif genes, a more compact representation of individuals. Our empirical study shows that the usage of motif genes improves coverage both for Evolutionary Algorithms and random approaches. In particular, it also shows that NSGA‐II, the multi‐objective evolutionary algorithm used by Sapienz, does not have a clear improvement over other algorithms. In terms of number of crashes detected, our study shows that both NSGA‐II and Random Search perform similarly. While the usage of motif genes improves the crash detection of algorithms, it is not enough to make it statistically significant. These facts cast doubts about the use of Evolutionary Algorithms in the context of Android test generation and suggest that motif genes can have a great impact on the overall effectiveness.

The 2020 global stock market crash: Endogenous or exogenous?

Starting on February 20, 2020, the global stock markets began to suffer the worst decline since the Great Recession in 2008, and the COVID-19 has been widely blamed on the stock market crashes. In this study, we applied the log-periodic power law singularity (LPPLS) methodology based on multilevel time series to unravel the underlying mechanisms of the 2020 global stock market crash by analyzing the trajectories of 10 major world stock market indexes from both developed and emergent stock markets, including the S&P 500, the DJIA, and the NASDAQ from the United State, the FTSE from the United Kingdom, the DAX from Germany, the NIKKEI from Japan, the CSI 300 from China, the HSI from Hong Kong, the BSESN from India, and the BOVESPA from Brazil. In order to effectively distinguish between endogenous crash and exogenous crash in stock market, we proposed using the LPPLS confidence indicator as a classification proxy. The results show that the apparent LPPLS bubble patterns of the super-exponential increase, corrected by the accelerating logarithm-periodic oscillations, have indeed presented in the price trajectories of the seven indexes: S&P 500, DJIA, NASDAQ, DAX, CSI 300, BSESN, and BOVESPA, indicating that the large positive bubbles have formed endogenously prior to the 2020 stock market crash, and the subsequent crashes for the seven indexes are endogenous, stemming from the increasingly systemic instability of the stock markets inherently, while the well-known external shocks, such as the COVID-19 pandemic, the corporate debt bubble, and the 2020 Russia–Saudi Arabia oil price war, only served as sparks during the 2020 global stock market crash. In contrast, the crashes in the three remaining indexes: FTSE, NIKKEI, and HSI, are exogenous and hence are perhaps the only crashes truly due to the COVID-19 pandemic. We also found that in terms of the regime changes of the stock markets, no obvious LPPLS negative bubble pattern has been observed in the price trajectories of the 10 stock market indexes, indicating that the regime changes from a bear market to a bull market in late March 2020 are exogenous, stemming from external factors. The unprecedented market and economy rescue efforts from federal reserves and central banks across the world in unison may have played a critical role in quelling the 2020 global stock market crash in the nick of time. This paper creates a paradigm for future studies in real-time crash detection and underlying mechanism dissection. It serves to warn us of the imminent risks in not only the stock market but also other financial markets and economic indexes.

Cyber-physical universal safety and crash detection system for autonomous robot

Multi-purpose mobile robots with a modular layout have become a hot research topic in recent years. The safe performance of these robot’s operations is relying on the low-level cyber-physical system (CPS). In this paper, the scientific goals underscore the analysis of the computational (cyber) units for low-level real-time fault monitoring. The purpose of these units is to monitor control signals issued by other (cyber) units or sensors and if there is a safety-critical problem, then predefined actions can be triggered. The safety controller was built and tested on the TalTech iseAuto platform. Based on the results, a new multi-layer universal safety system was developed. In the first layer, all signals and messages are checked to be in a suitable range or order. The second layer detects if an accident is happening using crash sensors. The third layer includes remote control switches support. If a failure or malfunction occurs, the emergency action plan is executed to stop the vehicle safely.

The ‘COVID’ crash of the 2020 U.S. Stock market

We employed the log-periodic power law singularity (LPPLS) methodology to systematically investigate the 2020 stock market crash in the U.S. equities sectors with different levels of total market capitalizations through four major U.S. stock market indexes, including the Wilshire 5000 Total Market index, the S&P 500 index, the S&P MidCap 400 index, and the Russell 2000 index, representing the stocks overall, the large capitalization stocks, the middle capitalization stocks and the small capitalization stocks, respectively. During the 2020 U.S. stock market crash, all four indexes lost more than a third of their values within five weeks, while both the middle capitalization stocks and the small capitalization stocks have suffered much greater losses than the large capitalization stocks and stocks overall. Our results indicate that the price trajectories of these four stock market indexes prior to the 2020 stock market crash have clearly featured the obvious LPPLS bubble pattern and were indeed in a positive bubble regime. Contrary to the popular belief that the 2020 US stock market crash was mainly due to the COVID-19 pandemic, we have shown that COVID merely served as sparks and the 2020 U.S. stock market crash had stemmed from the increasingly systemic instability of the stock market itself. We also performed the complementary post-mortem analysis of the 2020 U.S. stock market crash. Our analyses indicate that the probability density distributions of the critical time for these four indexes are positively skewed; the 2020 U.S. stock market crash originated from a bubble that had begun to form as early as September 2018; and the bubble profiles for stocks with different levels of total market capitalizations have distinct temporal patterns. This study not only sheds new light on the makings of the 2020 U.S. stock market crash but also creates a novel pipeline for future real-time crash detection and mechanism dissection of any financial market and/or economic index.

Car crash detection using ensemble deep learning and multimodal data from dashboard cameras

Due to the increase in motor vehicle accidents, there is a growing need for high-performance car crash detection systems. The authors of this research propose a car crash detection system that uses both video data and audio data from dashboard cameras in order to improve car crash detection performance. While most existing car crash detection systems depend on single modal data (i.e., video data or audio data only), the proposed car crash detection system uses an ensemble deep learning model based on multimodal data (i.e., both video and audio data), because different types of data extracted from one information source (e.g., dashboard cameras) can be regarded as different views of the same source. These different views complement one another and improve detection performance, because one view may have information that the other view does not contain. In this research, deep learning techniques, gated recurrent unit (GRU) and convolutional neural network (CNN), are used to develop a car crash detection system. A weighted average ensemble is used as an ensemble technique. The proposed car crash detection system, which is based on multiple classifiers that use both video and audio data from dashboard cameras, is validated using a comparison with single classifiers that use video data or audio data only. Car accident YouTube clips are used to validate this research. The experimental results indicate that the proposed car crash detection system performs significantly better than single classifiers. It is expected that the proposed car crash detection system can be used as part of an emergency road call service that recognizes traffic accidents automatically and allows immediate rescue after transmission to emergency recovery agencies.

A low cost and user friendly vehicle crash alert system using arduino

This paper is aimed at developing a vehicle crash detection system using Arduino, GPS, GSM and accelerometer. Accelerometer detects the sudden change in the axes of vehicle and GSM module sends the alert message on your mobile phone with the location of the accident. Location of accident is sent in the form of Google map link, derived from the latitude and longitude from GPS module. The message also contains the speed of vehicle in kilometers per hour. The proposed model can also be used as a tracking system and much more, by just making few changes in hardware and software. The alert system has been developed in real time and performance is discussed.

Exposure of Crash in Metal Foil Using UWB Antenna

Crash in aluminium are one of the earliest indications of degradation of products which are made up of aluminium. Detecting crashs in aluminum is a challenging task. Several non-destructive techniques have been used to detect the crack such as laser spot thermography, pulse eddy current technique, dye penetrant testing. In this paper, Ultra-wideband antenna is proposed for crash detection in aluminum. The Proposed Ultra-Wide Band (UWB) antenna has the capability of operating between 3.1GHz to 10.6GHz. The proposed antenna consists of rectangular radiating patch above the FR4 substrate which is having the dielectric constant 4.3 and thickness of 1.6mm. The results of the designed antenna like gain, radiation pattern, directivity, return loss is simulated by CST Microwave Studio software. The return loss features of the antenna shows the presents of crack in the aluminum.

A Novel Crash Detection Algorithm for Two-Wheeled Vehicles

This paper presents a crash detection strategy for motorcycles, using GPS/GNSS and inertial measurements collected by telematic e-Boxes. The primary goal is to jointly minimize the access time of the emergency services and to accurately store the event dynamics for further investigation on accident's responsibilities. For motorcycles, unlike for cars, crash events cannot always be detected by monitoring abnormal longitudinal decelerations solely. Thus, this work proposes a novel method that detects and classifies the severity of crash-like situations. The proposed approach follows a new paradigm that improves the detection performance, better generalizing the motorcycle dynamics with respect to the specific vehicle and the driving style. The proposed approach has been tested and validated with experimental data, covering both motorsport and naturalistic scenarions, involving several riders, different road conditions and vehicles.

An Empirical Analysis of Test Input Generation Tools for Android Apps through a Sequence of Events

Graphical User Interface (GUI) testing of Android apps has gained considerable interest from the industries and research community due to its excellent capability to verify the operational requirements of GUI components. To date, most of the existing GUI testing tools for Android apps are capable of generating test inputs by using different approaches and improve the Android apps’ code coverage and fault detection performance. Many previous studies have evaluated the code coverage and crash detection performances of GUI testing tools in the literature. However, very few studies have investigated the effectiveness of the test input generation tools, especially in the events sequence length of the overall test coverage and crash detection. The event sequence length generally shows the number of steps required by the test input generation tools to detect a crash. It is critical to highlight its effectiveness due to its significant effects on time, testing effort, and computational cost. Thus, this study evaluated the effectiveness of six test input generation tools for Android apps that support the system events generation on 50 Android apps. The generation tools were evaluated and compared based on the activity coverage, method coverage, and capability in detecting crashes. Through a critical analysis of the results, this study identifies the diversity and similarity of test input generation tools for Android apps to provide a clear picture of the current state of the art. The results revealed that a long events sequence performed better than a shorter events sequence. However, a long events sequence led to a minor positive effect on the coverage and crash detection. Moreover, the study showed that the tools achieved less than 40% of the method coverage and 67% of the activity coverage.

A New Video-Based Crash Detection Method: Balancing Speed and Accuracy Using a Feature Fusion Deep Learning Framework

Quick and accurate crash detection is important for saving lives and improved traffic incident management. In this paper, a feature fusion-based deep learning framework was developed for video-based urban traffic crash detection task, aiming at achieving a balance between detection speed and accuracy with limited computing resource. In this framework, a residual neural network (ResNet) combined with attention modules was proposed to extract crash-related appearance features from urban traffic videos (i.e., a crash appearance feature extractor), which were further fed to a spatiotemporal feature fusion model, Conv-LSTM (Convolutional Long Short-Term Memory), to simultaneously capture appearance (static) and motion (dynamic) crash features. The proposed model was trained by a set of video clips covering 330 crash and 342 noncrash events. In general, the proposed model achieved an accuracy of 87.78% on the testing dataset and an acceptable detection speed (FPS > 30 with GTX 1060). Thanks to the attention module, the proposed model can capture the localized appearance features (e.g., vehicle damage and pedestrian fallen-off) of crashes better than conventional convolutional neural networks. The Conv-LSTM module outperformed conventional LSTM in terms of capturing motion features of crashes, such as the roadway congestion and pedestrians gathering after crashes. Compared to traditional motion-based crash detection model, the proposed model achieved higher detection accuracy. Moreover, it could detect crashes much faster than other feature fusion-based models (e.g., C3D). The results show that the proposed model is a promising video-based urban traffic crash detection algorithm that could be used in practice in the future.

Smart driver assistance system using raspberry pi and sensor networks

With the evolution of science and technology, monitoring human reactions and activities have become really easy and smooth. These new technologies have the potential to revolutionize the domain of safety and security in different realms of the society. Surveillance being the key factor of security measures has been elevated to a whole new level with the advancement in signal processing techniques. This paper basically focuses on the implementation of a smart surveillance system using signal processing and embedded tools which is applied in automobiles to ultimately develop the holistic driver assistance system. Earlier methods were based on physiological and analog data, but the present day scenario demands a smarter and digitalized working system so as to employ integrity and compatibility with other smart sub-systems like mobile phones and tablets. Transportation as we all know is one of the key sectors in the society. But the safety and security measures which people implement for their homes is not being employed for their vehicles. Apart from the vehicular anti-theft burglar systems, driver monitoring systems are also crucial to the lives of the driver and the passengers. Hence, this paper consists of three inter-linked modules which are the driver fatigue detection, alcohol content detection and vehicular crash detection along with control to monitor the driver's physiological state that can affect the vehicular control. A variety of input extraction hardware tools and software algorithms have been utilized in a collaborative way to implement this process.

Review on big data applications in safety research of intelligent transportation systems and connected/automated vehicles

The era of Big Data has arrived. Recently, under the environment of intelligent transportation systems (ITS) and connected/automated vehicles (CAV), Big Data has been applied in various fields in transportation including traffic safety. In this study, we review recent research studies that employed Big Data to analyze traffic safety under the environment of ITS and CAV. The particular topics include crash detection or prediction, discovery of contributing factors to crashes, driving behavior analysis, crash hotspot identification, etc. From the reviewed studies, employing advanced analytics for Big Data has a great potential for understanding and enhancing traffic safety. Big Data application in traffic safety integrates and processes massive multi-source data, breaks through the limitations of the traditional data analytics, and discovers and solves the problems, which cannot be solved by the traditional safety analytics. Lastly, suggestions are provided for future Big Data safety analytics under the environment of ITS and CAV.

Method for Classification of Frontal Collision Events in Passenger Cars Based on Measurement of Local Component-Specific Decelerations

The detection of accident scenarios is essential for a timely deployment of restraint devices and therefore for optimum protection of the vehicle occupants. Based on an innovative concept for crash detection, which involves measuring component-related local decelerations, this paper presents an entirely new method for the simulation and evaluation and estimates this with of a comprehensive set of crash load cases. With this approach, decelerations are detected directly at numerous individual components in the vehicle front end and are integrated in a velocity reduction using small time intervals. An evaluation based on multivariate statistical methods shows that the information content which results from exceedance of one defined velocity reduction threshold per measuring point is sufficient to safely distinguish between and classify all relevant load cases with a high level of independence. The concept has therefore proven to be functional and will be transferred to initial test series.

Life Case: Inelastic Collision based Automobile Crash Detection and Alert System via 3G Network

Accident detection and alert systems that can pinpoint the whereabouts of an accident are crucial to ensure the concerned authorities are informed instantaneously about the occurrence of an accident, in order for the deployment of emergency response to save lives in the least amount of time. A 3G incorporated accident detection system, known as Life Case is designed to discover the accidents between automobiles and, send the timestamp, and the actual position of an accident, to a developed android application built using Android Studio, via a cloud database. The alert message is presented using Google Maps which helps user to save plenty of time through easy navigation feature towards the accident location. Life Case measures the acceleration of an automobile during a collision using the Theory of Inelastic Collision.

A long short-term memory-based framework for crash detection on freeways with traffic data of different temporal resolutions

Traffic crash detection is a major component of intelligent transportation systems. It can explore inner relationships between traffic conditions and crash risk, prevent potential crashes, and improve road safety. However, there exist some limitations in current studies on crash detection: (1) The commonly used machine learning methods cannot simulate the evolving transitions of traffic conditions before crash occurrences; (2) Current models collected traffic data of only one temporal resolution, which cannot fully represent traffic trends in different time intervals. Therefore, this study proposes a Long short-term memory (LSTM) based framework considering traffic data of different temporal resolutions (LSTMDTR) for crash detection. LSTM is an effective deep learning method to capture the long-term dependency and dynamic transitions of pre-crash conditions. Three LSTM networks considering traffic data of different temporal resolutions are constructed, which can comprehensively indicate traffic variations in different time intervals. A fully-connected layer is used to combine the outputs of three LSTM networks, and a dropout layer is used to reduce overfitting and improve prediction performance. The LSTMDTR model is implemented on datasets of I880-N and I805-N in California, America. The results indicate that the LSTMDTR model can obtain satisfactory performance on crash detection, with the highest crash accuracy of 70.43 %. LSTMDTR models constructed on one freeway can be transferred to other similar freeways, with 65.12 % of crash accuracy on transferability. Compared with machine learning methods and LSTM models with one or two temporal resolutions, the LSTMDTR model has been validated to perform better on crash detection and transferability. A proper number of neurons in the LSTMDTR model should be determined in real applications considering acceptable detection performance and computation time. The dropout technique can reduce overfitting and improve the generalization ability of the LSTMDTR model, increasing crash accuracy from 64.49 % to 70.43 %.

MERA (motorcycle emergency ride assistance): plug-in for accident detection and notification

The condition to trigger the Automatic crash detection for motorcycles is more complex, since the collisions in real life are usually different than those in the lab. so in order to boost the accuracy and affectivity, the researchers must reduce inconsequential deployments, the occupant weight, and riding condition, as well as re-calculate the speed to crash conditions. In this paper was present a continuously evaluates and innovate for crash detection with support automated notification system. The accident detection, communicates the accelerometer, vibration, speed and lean values to the determine accident status and level. In this concept, three different variable measurements are attached to head of the motorist and motorcycle body, Crash dummy tests are done for trial particular systems in helmet by throwing the helmet with different altitudes to simulate the effect of crash to the motorist, and for capture the location of accident, real data is collected by driving the motorcycle. The implemented prototype system showed promising results for automatic crash detection, and the system also showed its success for deliver real-time notifications when an accident was happening.

A Vision-Based Video Crash Detection Framework for Mixed Traffic Flow Environment Considering Low-Visibility Condition

In this paper, a vision-based crash detection framework was proposed to quickly detect various crash types in mixed traffic flow environment, considering low-visibility conditions. First, Retinex image enhancement algorithm was introduced to improve the quality of images, collected under low-visibility conditions (e.g., heavy rainy days, foggy days and dark night with poor lights). Then, a Yolo v3 model was trained to detect multiple objects from images, including fallen pedestrians/cyclists, vehicle rollover, moving/stopped vehicles, moving/stopped cyclists/pedestrians, and so on. Then, a set of features were developed from the Yolo outputs, based on which a decision model was trained for crash detection. An experiment was conducted to validate the model framework. The results showed that the proposed framework achieved a high detection rate of 92.5%, with relatively low false alarm rate of 7.5%. There are some useful findings: (1) the proposed model outperformed empirical rule-based detection models; (2) image enhancement method can largely improve crash detection performance under low-visibility conditions; (3) the accuracy of object detection (e.g., bounding box prediction) can impact crash detection performance, especially for minor motor-vehicle crashes. Overall, the proposed framework can be considered as a promising tool for quick crash detection in mixed traffic flow environment under various visibility conditions. Some limitations are also discussed in the paper.

Process monitoring and impulse detection in face milling using capacitive acceleration sensors based on MEMS

High-performance acceleration sensors acquire increasing relevance for process monitoring and crash detection. In this regard, a newly developed sensor system is realised comprising an 8 kHz and a 20 kHz high-acceleration MEMS-based sensor. This research presents proof of functionality of the sensor system conducting cutting experiments and impulse detection tests. For cutting tests, a 42CrMo4 steel grade is machined with cemented carbide milling tools. The process data indicate that distinct cutting processes stages and crash related impulses can be recognised reliably. As general functionality is proven, the sensor system can be evolved for industrial process monitoring and crash detection applications.

Highway crash detection and risk estimation using deep learning

Crash Detection is essential in providing timely information to traffic management centers and the public to reduce its adverse effects. Prediction of crash risk is vital for avoiding secondary crashes and safeguarding highway traffic. For many years, researchers have explored several techniques for early and precise detection of crashes to aid in traffic incident management. With recent advancements in data collection techniques, abundant real-time traffic data is available for use. Big data infrastructure and machine learning algorithms can utilize this data to provide suitable solutions for the highway traffic safety system. This paper explores the feasibility of using deep learning models to detect crash occurrence and predict crash risk. Volume, Speed and Sensor Occupancy data collected from roadside radar sensors along Interstate 235 in Des Moines, IA is used for this study. This real-world traffic data is used to design feature set for the deep learning models for crash detection and crash risk prediction. The results show that a deep model has better crash detection performance and similar crash prediction performance than state of the art shallow models. Additionally, a sensitivity analysis was conducted for crash risk prediction using data 1-minute, 5-minutes and 10-minutes prior to crash occurrence. It was observed that is hard to predict the crash risk of a traffic condition, 10 min prior to a crash.

Advanced automobile crash detection by acoustic methods

Today’s automobiles are equipped with an increasing variety of safety features, yet the use of acoustic methods for automobile crash prevention and detection has been somewhat limited even though the acoustic waves generated during such events can offer valuable information. For example, the high-pitched squealing caused by tire skidding can provide advance warning especially if it is caused by an adjacent car. During car collisions, the elastic waves traveling along the steel car frame are 17 times faster than the speed of sound in air, which can signal a crash more promptly than center-mounted acceleration sensors. To make full use of the high-speed acoustic signals, a wavelet-based algorithm implementable in real-time has been developed to isolate and detect specific pre-crash and crash events such as honking, tire skidding and collision in multi-channel acoustic datasets. The proposed algorithm offers distinct advantages in sudden onset detection, temporal localization accuracy, and computational cost over existing time- and frequency-domain methods. Results demonstrated on a crash scenario are indicative of a substantial enhancement in automobile pre-crash and crash detection performance by acoustic methods.